Social: Theories and Models - Foundations for Designing User-Centered Systems - page 257

(Dunbar 1992 ). He related the maximum productive group size to the size of the
neocortex in primates, and in reference to strong social relationships, not distant or
distal ones. The number is related to the question ''how much time do you have to
maintain close social ties''? Some of his ideas do not fully apply to the online
world, and some of the data he presents seem anecdotal. Further writers and
researchers have had a hard time coming to grips with the question; for example, it
has been speculated that the scientific subspecialties can't be more than 150 sci-
entists because of this effect (but this then implies that either the scientists have no
other friends, or that they can handle exactly twice as many as other people!). How
do you measure relationship strength? What counts as a strong relationship? Does
improved communication media improve the ability to have a larger number of
ties? Nevertheless, it remains a useful question, and has implications for system
design, and limits are being applied in some systems.
What is perhaps most important from a system design point of view is the active
connections. In other words, the connections that are used rather than dormant. If
you think about Facebook, for example, some people will boast about having, say,
352 Facebook friends, but when you analyze how many people they really interact
with, the number is much smaller, by a factor of between 10 and 100. If all of the
Facebook friend connections were active, it would be a full time job for an
individual just to monitor their friend's ongoing activity.
The concept of the Dunbar number provides suggestions of how many social
connections systems should support. For example, how many email aliases should
you support in an email program, or how many names can be on an email alias?
What are optimal sized online groups? Can a classroom of 150 students (online or
in a building) be a group, or can it only be a group of groups?
9.2.5 Good Personal Social Networks Lead to Better Work 1
Burt ( 2004 ) examined how well the managers in an electronics company were
connected to other parts of the company. This work serves as a good example of what
implications there are for and from networks for designing teams and systems to
support them, as well as how social networks influence systems. Burt computed how
connectedness correlated with objective measures of job performance (i.e., salary
and promotions) and how well connectedness correlated with subjective measures of
performance (i.e., job evaluations and evaluations of their ideas by supervisors).
This is a type of network analysis, in that Burt examined how the managers
were connected to groups in the company, and to whom. A summary is shown in
Fig. 9.4 . He also used the term structural holes, which occur when two groups that
should be connected are not. For example, if two groups working on related,
1 Umer Farooq brought this work to our attention as part of a talk by Peter Pirolli. We thank
Umer and Peter for their help.